Homework

A simple dynamic web application in a CI/CD world.

Web App

Dependency Installation

Dependencies:

Flask Python web framework
the waitress production-level server.

Development specific:

Pythonic wheel packaging tool pbr
testing tools pytest and coverage

To install the dependencies and run the app, ensure you have the following tools installed on your host:

Python, version >= 3.7
pipenv, version >= 2018.11.26

Run:

pipenv install --dev

Development Usage

This will start a foreground server process, occupying the shell:

FLASK_APP=homework FLASK_ENV=development pipenv run flask run

Test the App

Black Box Testing

In a seprate shell invoke an ubiquitous HTTP client:

curl -i 'http://localhost:5000/dynamic'

Expect output similar to the below text:

HTTP/1.0 200 OK
Content-Type: text/html; charset=utf-8
Content-Length: 19
Server: Werkzeug/0.16.0 Python/3.7.6
Date: Sun, 19 Jan 2020 17:09:42 GMT

0.29421629165877494

Built-in Tests

Run:

pipenv run coverage run -m pytest
pipenv run coverage report

You should see similar output to the following:

======================= test session starts =========================
platform linux -- Python 3.7.6, pytest-5.3.3, py-1.8.1, pluggy-0.13.1
rootdir: [redacted]/homework, inifile: setup.cfg, testpaths: tests
collected 3 items                                                                                                 

tests/test_dynamic.py ..                                       [ 66%]
tests/test_factory.py .                                        [100%]

======================= 3 passed in 40.11s ==========================

Name                   Stmts   Miss Branch BrPart  Cover
--------------------------------------------------------
homework/__init__.py      21      0      2      0   100%
homework/dynamic.py        7      0      0      0   100%
--------------------------------------------------------
TOTAL                     28      0      2      0   100%

Production Usage

Run the waitress server:

pipenv run waitress-serve --listen '0.0.0.0:8080' \
                          --connection-limit=2000 \
                          --asyncore-use-poll '
                          --call '
                          'homework:create_app'

> booting
> booted
Serving on http://0.0.0.0:8080

Packaging Into a Wheel

Inspired by @greut. A wheel package can be built using:

pipenv run python setup.py bdist_wheel
tree dist/

dist/
└── homework-0.0.1.dev12-py3-none-any.whl

Containerization

Building an Image

Use your favourite build tool, e.g.:

docker build homework:latest .
podman build -t homework:latest --format docker .
buildah bud -t homework:latest --format docker .

Running an Image

docker run --rm -d -p 8080:8080 homework:latest
podman run --rm -d -p 8080:8080 homework:latest

Test with an ubiquitous HTTP client like mentioned before.

Cloudification

For some reason (gratis 300$ debit) Google Cloud was chosen to host the application.

Not everything on GCloud was automated:

registering account
enabling GKE service
creating a project
creating a service account for the project and issuing a deployment key for it

Provision K8S Cluster

Automated using Ansible, but not part of CI. When Ansible is used against APIs, it usually runs on localhost where some dependencies for the Ansible modules are installed. Again, make sure these basic tools exist:

Python, version >= 3.7
pipenv, version >= 2018.11.26

Using Ansible, connect to Google Cloud and provision a container/GKE cluster and a node pool. Make sure environment variables contain proper values.

Warning: Many properties cannot be changed after cluster creation.

pushd deploy/provision/
python3 -m pip install pipenv
pythin3 -m pipenv
GCP_AUTH_KIND=serviceaccount \
GCP_SA_KEY_FILE=/path/to/gcp_serviceaccount_credentials_file.json \
GCP_LOCATION=europe-west3-c \
GCP_PROJECT=x-ripple-123456 \
pipenv run ansible-playbook gcloud-ansible-playbook.yml
popd

Apply K8S Deployment and Service Declarations

Some tools are required to interface with Google Cloud and the Kubernetes cluster. Here's a shell script for installation using the basic tools bash and curl:

curl -O https://dl.google.com/dl/cloudsdk/channels/rapid/downloads/google-cloud-sdk-277.0.0-linux-x86_64.tar.gz
test $(sha256sum google-cloud-sdk-277.0.0-linux-x86_64.tar.gz | cut -d\  -f1) == b158894c427712d006fac235daf8d288db110ecfbcf649ed11cdf36ac2b2ff2a || exit
tar xzf google-cloud-sdk-277.0.0-linux-x86_64.tar.gz
yes | bash google-cloud-sdk/install.sh
source google-cloud-sdk/path.bash.inc
yes | gcloud components update
yes | gcloud components install kubectl

Now create a kubectl configuration so it can auhtentication against the created cluster. Some varible names are used below, see their definitionin the previous chapter.

gcloud auth activate-service-account --key-file=$GCP_SA_KEY_FILE
gcloud container clusters get-credentials homework-1-gke-cluster --project=$GCP_PROJECT --zone=$GCP_LOCATION

Now the workload on the GKE cluser can finally be deployed and exposed:

cat deployment.yml service.yml | kubectl apply -f-

Performance testing

The workload on GKE is exposed on by a Load Balancer IP. It can then receive requests. To test performance, one can use a variety of tools, e.g. ab, wrk, JMeter, tsung. Since this homework application is very simple and has only a few resources, invoking the simple wrk for each is enough:

for url in http://${gke_service_ip}/{,dynamic}
do
    wrk -c 400 -d 60 --latency $url
done

Output:

Running 1m test @ http://35.234.75.59/
  2 threads and 400 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency   313.29ms  107.43ms   1.97s    76.17%
    Req/Sec   604.71    381.79     1.98k    67.04%
  Latency Distribution
     50%  304.33ms
     75%  364.38ms
     90%  433.84ms
     99%  607.68ms
  71735 requests in 1.00m, 9.99MB read
  Socket errors: connect 0, read 0, write 0, timeout 203
Requests/sec:   1193.68
Transfer/sec:    170.19KB

Running 1m test @ http://35.234.75.59/dynamic
  2 threads and 400 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency   300.49ms  107.46ms   1.98s    76.39%
    Req/Sec   649.85    428.48     2.05k    65.15%
  Latency Distribution
     50%  290.85ms
     75%  349.85ms
     90%  423.84ms
     99%  608.31ms
  76424 requests in 1.00m, 11.61MB read
  Socket errors: connect 0, read 0, write 0, timeout 145
Requests/sec:   1271.91
Transfer/sec:    197.83KB

bartmanus/homework-1